This project is to apply catalytic model to fit age-stratified cases data to estimate the Force of infection (FOI) distribution of Japanese Encephalitis in endemic area (VIMC supported countries). Then using the vaccine coverage data to generate burden as given template. This project is created by R programming language.
The project is organized as following:
- Script: Entire code files for the project
- Data folder: Contains data that is necessary to run the script (Please contact the author to get the Data folder)
- Generate folder: Contains the result after running the script
- Shiny folder: A Shiny application which can visualize and compare the generated burden csv file from Step 8. This application requires an user to upload 2 burden templates, then it will plot the burden values of 2 templates and show the difference between them.
Need to install the following libraries: rstan, readxl, shiny, ggplot2, gridExtra (the last 3 libraries are used in Shiny application)
- Run Rstan to fit the catalytic model to age-stratified cases data (Go to Step 1)
- Create Susceptible population for the Naive (No vaccination) scenario (Go to Step 2)
- Create Susceptible population for the Routine scenario (Go to Step 3)
- Create Susceptible population for the Campaign scenario (Go to Step 4)
- Create Burden (Cases, Deaths, DALYs) for each scenario (MeanBurden) (Go to Step 5)
- (Optional) Create Burden (Cases, Deaths, DALYs) for each scenario (MeanFOI) (Go to Step 6)
- Create Burden STOCHASTIC (Cases, Deaths, DALYs) for each scenario (Go to Step 7)
- Fill the Burden result to given template (Go to Step 8)
- Fill the Burden STOCHASTIC result to the STOCHASTIC template (Go to Step 9)
Create (or run the given) Rstan model to fit the catalytic model to the given age-stratified cases data. Note that there are 2 regions that we do not have age-stratified cases data, which are PAK and IND.Low. For these 2 regions, we sampled with a log-norm distribution with mean = 0.01 and sd = 1.
Input
- Age-stratified cases data in xls format is stored in Data/Age_Stratified_Cases_Data/ folder.
- Rstan model script (if already given) is stored in Generate/Rstan_Model/ folder.
Output
- The stan result for each regions, named FOI_distribution_[Region].Rds, will be stored in Generate/FOI_Distribution/ folder. This result will not only contain FOI distribution (as lambda component in the result), but also other information of stan.
- Rstan model script (if there is not a model in Generate/Rstan_Model/ folder) will be stored in that folder.
Function
- Run_Stan_FOI: Run this script to fit the model
Create Susceptible population at each age in csv file in the Naive (no vaccination) scenario. The susceptible population in this scenario is quite similar with the original population given by VIMC. However, for some special countries (like CHN, PAK, ...) in which the endemic area are subnational regions. In these cases we need to adjust the population by using population at risk information from WHO or previous studies. This script also provided the option to save the susceptible population by countries seperately.
Input
- Population data given by VIMC is stored in Data/Population/ folder.
- Population at risk information is writen in the script
Output
- NaivePop.csv is produced in Generate/Susceptible_Population/ folder.
- Susceptible population for each country NaivePop_[ISO].csv are produced in the subfoler Generate/Susceptible_Population/Countries. (If you choose the option to save seperately)
Function
- Create_Naive_Pop: Create susceptible population in the naive scenario
Update 24 Aug 2019: Now you can choose the option portion_vaccinated_run run. If this is set to TRUE, the vaccinated people will portionally grow up (only some vaccinated people last year will be still alive in the next year). If this is set to FALSE, all vaccinated people in the last year will survive in the next year (no death)
Create Susceptible population at each age in csv file in the Routine scenario. Routine scenario means only people at age 0 will receive vaccination. The portion of people that will be vaccinated is described in the vaccine coverage file from VIMC. The procedure of this step is to calculate number of vaccinated people in routine scenario firstly. Then find the remaining susceptible population. This will equal naive population (susceptible population when there is no vaccination) from Step 2 minus number of vaccinated people in routine scenario.
Input
- Vaccine coverage given by VIMC is stored in Data/Vaccine_Coverage/ folder.
- Naive population produced in Step 2 (stored in Generate/Susceptible_Population/ folder).
Output
- RoutinePop.csv, indicating susceptible population, is produced in Generate/Susceptible_Population/ folder.
- RoutineVac.csv, indicating vaccinated population, is produced in Generate/Vaccinated_Population/ folder.
- Susceptible population for each country RoutinePop_[ISO].csv are produced in the subfoler Generate/Susceptible_Population/Countries. (If you choose the option to save seperately)
- Vaccinated population for each country RoutineVac_[ISO].csv are produced in the subfoler Generate/Vaccinated_Population/Countries. (If you choose the option to save seperately)
Function
- Create_Routine_Pop: Create susceptible population in the routine scenario (after the vaccination)
Update 24 Aug 2019: Now you can choose the option portion_vaccinated_run run. If this is set to TRUE, the vaccinated people will portionally grow up (only some vaccinated people last year will be still alive in the next year). If this is set to FALSE, all vaccinated people in the last year will survive in the next year (no death)
Create Susceptible population at each age in csv file in the Campaign scenario. Campaign scenario means they will conduct Routine vaccination first, then perform a massive vaccination for the remaining susceptible people in the selected age range (normally from age 0 to age 14). The portion of people that will be vaccinated is described in the vaccine coverage file from VIMC. The procedure of this step is to calculate number of vaccinated people in routine scenario firstly. Then find the remaining susceptible population and calculate the vaccinated people in the campaign scenario. Finally we will the total remaining susceptible people, which will equal naive population (susceptible population when there is no vaccination) from Step 2 minus number of vaccinated people found as above.
Input
- Vaccine coverage given by VIMC is stored in Data/Vaccine_Coverage/ folder.
- Naive population produced in Step 2 (stored in Generate/Susceptible_Population/ folder).
Output
- CampaignPop.csv, indicating susceptible population, is produced in Generate/Susceptible_Population/ folder.
- CampaignVac.csv, indicating vaccinated population, is produced in Generate/Vaccinated_Population/ folder.
- Susceptible population for each country CampaignPop_[ISO].csv are produced in the subfoler Generate/Susceptible_Population/Countries folder. (If you choose the option to save seperately)
- Vaccinated population for each country CampaignVac_[ISO].csv are produced in the subfoler Generate/Vaccinated_Population/Countries folder. (If you choose the option to save seperately)
Function
- Create_Campaign_Pop: Create susceptible population in the campaign scenario (after the vaccination)
Update 26 Aug 2019: We now also calculated the immuned population (asymptomatic and symptomatic cases). Cases = Immuned * Symptomatic Rate
This step will calculate Cases, Deaths, and DALYs for each scenario of vaccination. We will use the FOI distribution from Step 1, the susceptible population for each scenario from Step 2, Step 3, Step 4. Besides, we also need the information of Life Expactation, which is provided by VIMC. Because the FOI distribution includes many values due to the simulation from Rstan, here we calculate burden based on each value, then take the mean all of them to have the final burden.
Input
- NaivePop.csv, RoutinePop.csv, CampaignPop.csv: Susceptible population for each vaccination scenario from Step 2, Step 3, Step 4.
- FOI_distribution from Step 1.
- Life Expectancy information stored in Data/Life_Expectation/ folder.
Output
- Naive/Routine/Campaign_Burden_MeanBurden.Rds: 3 lists of lists containing calculated burden for each scenario will be saved in Generate/Burden/
Function
- Create_Burden_MeanBurden: Create burden lists for each scenario. Each list describes one scenario and it includes 3 sub-lists for cases, deaths, and dalys
Update 26 Aug 2019: We now also calculated the immuned population (asymptomatic and symptomatic cases). Cases = Immuned * Symptomatic Rate
Basically, this step is optional and quite similar with Step 5. Instead calculating burden for each of all FOI values, here we take the mean of FOI distribution firstly, then calculate burden based on that average FOI value.
Input
- NaivePop.csv, RoutinePop.csv, CampaignPop.csv: Susceptible population for each vaccination scenario from Step 2, Step 3, Step 4.
- FOI_distribution from Step 1.
- Life Expectancy information stored in Data/Life_Expectation/ folder.
Output
- Naive/Routine/Campaign_Burden_MeanFOI.Rds: 3 lists of lists containing calculated burden for each scenario will be saved in Generate/Burden/
Function
- Create_Burden_MeanFOI: Create burden lists for each scenario. Each list describes one scenario and it includes 3 sub-lists for cases, deaths, and dalys
Update 26 Aug 2019: We now also calculated the immuned population (asymptomatic and symptomatic cases). Cases = Immuned * Symptomatic Rate
Basically, this step is optional and quite similar with Step 5. Instead taking the mean of all calculated burden, here we sample some of them (default is to randomly choose 200 from 1600 burden values)
Input
- NaivePop.csv, RoutinePop.csv, CampaignPop.csv: Susceptible population for each vaccination scenario from Step 2, Step 3, Step 4.
- FOI_distribution from Step 1.
- Life Expectancy information stored in Data/Life_Expectation/
Output
- Naive/Routine/Campaign_Burden_Stochastic.Rds: 3 lists of lists containing calculated burden for each scenario will be saved in Generate/Burden_Stochastic/. Each list will include 200 (default) lists for each run.
Function
- Create_Burden_Stochastic: Create burden lists for each scenario in Stochastic run.
Update 26 Aug 2019: Cohort size now will be equal to Susceptible + Vaccinated + Immuned = NaivePop + Immuned. The last 'equal sign' is because Naive = Susceptible + Vaccinated
Fill in the template given by VIMC with burden (cases, deaths, dalys) and cohort size.
Input
- Naive/Routine/Campaign_Burden.Rds: Burden list found from Step 5 (or Step 6)
- Burden template, given by VIMC, is stored at Data/Burden_Template/
- Population data, given by VIMC, is stored at Data/Population/
- Susceptible population in Naive scenario (from Step 2)
Output
- Template_Naive/Routine/Campaign.csv: 3 filled csv files will be stored at Generate/Template/[Scenario]/
Function
- Fill_Template: Fill the burden and cohort size to the given template (from VIMC)
Update 26 Aug 2019: Cohort size now will be equal to Susceptible + Vaccinated + Immuned = NaivePop + Immuned. The last 'equal sign' is because Naive = Susceptible + Vaccinated
Fill in the stochasti template given by VIMC with burden (cases, deaths, dalys) and cohort size in stochastic run. Basically, this step is almost the same as Step 8. However, in stochastic run, we need to create a CSV file that stores all the parameters for each run id (Such as symptomatic rate, mortality rate for each run id). Besides, we also produce many burden csv files (one for each run id and for each scenario)
Input
- Naive/Routine/Campaign_Burden_Stochastic.Rds: Burden list found from Step 7
- Burden STOCHASTIC template, given by VIMC, is stored at Data/Burden_Template/
- Population data, given by VIMC, is stored at Data/Population/
- Susceptible population in Naive scenario (from Step 2)
Output
- Template_Stochastic_Naive/Routine/Campaign_[runid].csv: series of CSV files will be stored at Generate/Template_Stochastic/[Scenario] folder.
- Template_Stochastic_Parameters.csv: CSV files that contains value of parameters used to generate burden for each run.
Function
- Fill_Template_Stochastic: Fill the burden and cohort size to the given template (from VIMC)
This part is optional and not really important.
- Change_stochastic_template_name:Change the result stochastic template name (Step 9) to match with the rule given by VIMC
- Convert_Format_Original_Pop: Convert format of population data from new format (run 201710gavi-6) to old format (which will be used in entire the project). The examples of 2 new and old format are stored in Data/Population_Format/. Note that this script might not work well if the newest format is different with the given 201710gavi-6 format.